[Pw_forum] Interpreting starting_ns_eigenvalue in the output
Cococcioni Matteo
matteo.cococcioni at epfl.ch
Fri Feb 2 10:40:51 CET 2018
Dear Kayahan,
I might be wrong (I cannot check right now, and the code might have changed in that part) but I seem to remember that by fixing occupations from input you only change the eigenvalues of the occupation matrix, not the eigenvectors. In other words, you are not forcing occupation 1 on the zy and xy states but on the third and fifth eigenstates of the occupation matrix (putting all other to zero). This is done after the first iteration. So, in order to understand how you want to modify it, you have to look at the occupation matrix for the species of interest after the first diagonalization.
Hope this helps.
Regards,
Matteo
On Feb 2, 2018, at 6:52 AM, Kayahan Saritas <saritaskayahan at gmail.com<mailto:saritaskayahan at gmail.com>> wrote:
Hi,
I have been using the starting_ns_eigenvalue keyword in order to force the system a specific occupation of orbitals at the initial steps of scf optimization. This is for the QE-5.3.0 (I know it is old version, I have to use this one for compatibility issues with a quantum Monte Carlo code, so some of the settings in my input may look unusual for the same reason). My input is the following below the post, it is basically AFM-II structure for rocksalt CoO, it has a some rhombohedral distortion that is observed around 0K.
I am following the starting_ns_eigenvalue(m,ispin,l) notation in pw.x manual, for example starting_ns_eigenvalue(1,1,1)=1 would mean that for species 1, majority spin in dz2 orbital, occupation is set to 1. I borrowed the order of m values from projwfc.x manual. So in the input, I intend to set occupations at dzy and dxy orbitals equal to one for species 1, minority spin (and species 2, majority spin). However, when I read the output, I get the following order (given below) during the first scf step. According to that 4th and 5th orbitals have an occupation of 1, not the 3rd and 5th that I asked in the input. Is it that QE grouping occupied and unoccupied orbitals together such that the order doesn’t matter or am I making a mistake in giving the right setup at the input. I would appreciate your feedback.
Thanks,
Kayahan
ORNL
———————————INPUT—————————————
&CONTROL
calculation = 'scf'
outdir = 'pwscf_output'
prefix = 'pwscf'
pseudo_dir = './'
/
&SYSTEM
celldm(1) = 1.0
degauss = 0.0005
ecutwfc = 350
Hubbard_U(1) = 5
Hubbard_U(2) = 5
ibrav = 0
input_dft = 'LDA'
lda_plus_u = .true.
nat = 4
nosym = .false.
nspin = 2
ntyp = 3
occupations = 'tetrahedra'
starting_ns_eigenvalue(1,1,1)=0
starting_ns_eigenvalue(2,1,1)=0
starting_ns_eigenvalue(3,1,1)=1
starting_ns_eigenvalue(4,1,1)=0
starting_ns_eigenvalue(5,1,1)=1
starting_ns_eigenvalue(1,2,1)=1
starting_ns_eigenvalue(2,2,1)=1
starting_ns_eigenvalue(3,2,1)=1
starting_ns_eigenvalue(4,2,1)=1
starting_ns_eigenvalue(5,2,1)=1
starting_ns_eigenvalue(1,1,2)=1
starting_ns_eigenvalue(2,1,2)=1
starting_ns_eigenvalue(3,1,2)=1
starting_ns_eigenvalue(4,1,2)=1
starting_ns_eigenvalue(5,1,2)=1
starting_ns_eigenvalue(1,2,2)=0
starting_ns_eigenvalue(2,2,2)=0
starting_ns_eigenvalue(3,2,2)=1
starting_ns_eigenvalue(4,2,2)=0
starting_ns_eigenvalue(5,2,2)=1
tot_charge = 0
tot_magnetization = 0
/
&ELECTRONS
conv_thr = 1e-07
electron_maxstep = 500
mixing_beta = 0.7
mixing_fixed_ns = 8
/
ATOMIC_SPECIES
Co1 58.933 Co.opt.upf
Co2 58.933 Co.opt.upf
O 15.999 O.opt.upf
ATOMIC_POSITIONS alat
Co1 -5.65718220 0.00578461 0.01409542
Co2 -5.64220976 -3.25742414 -4.63419238
O -8.49505366 -4.90328622 -6.94947095
O -2.78938403 -1.61155487 -2.31897499
K_POINTS automatic
4 4 4 0 0 0
CELL_PARAMETERS cubic
-2.82359317 -4.90156048 0.01386135
-5.65717990 0.00578508 0.01412863
-2.80366009 -1.61906839 -9.29643767
———————————OUTPUT—————————————
Modify starting ns matrices according to input values
--- enter write_ns ---
LDA+U parameters:
U( 1) = 5.00000000
alpha( 1) = 0.00000000
U( 2) = 5.00000000
alpha( 2) = 0.00000000
atom 1 Tr[ns(na)] (up, down, total) = 2.00000 5.00000 7.00000
spin 1
eigenvalues:
-0.000 -0.000 0.000 1.000 1.000
eigenvectors:
0.000 0.005 0.001 0.003 0.990
0.157 0.589 0.003 0.246 0.005
0.244 0.029 0.681 0.043 0.004
0.563 0.020 0.314 0.103 0.001
0.036 0.357 0.002 0.605 0.000
occupations:
0.994 -0.044 -0.049 -0.007 0.044
-0.044 0.251 0.107 0.161 0.386
-0.049 0.107 0.046 0.068 0.161
-0.007 0.161 0.068 0.104 0.250
0.044 0.386 0.161 0.250 0.605
spin 2
eigenvalues:
1.000 1.000 1.000 1.000 1.000
eigenvectors:
0.000 0.001 0.990 0.002 0.008
0.124 0.140 0.006 0.475 0.255
0.584 0.009 0.004 0.015 0.388
0.200 0.367 0.000 0.104 0.328
0.091 0.485 0.000 0.404 0.020
occupations:
1.000 -0.000 -0.000 -0.000 -0.000
-0.000 1.000 0.000 0.000 -0.000
-0.000 0.000 1.000 0.000 0.000
-0.000 0.000 0.000 1.000 0.000
-0.000 -0.000 0.000 0.000 1.000
atomic mag. moment = -3.000000
atom 2 Tr[ns(na)] (up, down, total) = 5.00000 2.00000 7.00000
spin 1
eigenvalues:
1.000 1.000 1.000 1.000 1.000
eigenvectors:
0.004 0.010 0.638 0.022 0.326
0.486 0.188 0.078 0.237 0.012
0.043 0.014 0.282 0.006 0.654
0.169 0.089 0.001 0.734 0.007
0.298 0.699 0.001 0.001 0.001
occupations:
1.000 0.000 -0.000 -0.000 0.000
0.000 1.000 -0.000 -0.000 -0.000
-0.000 -0.000 1.000 -0.000 -0.000
-0.000 -0.000 -0.000 1.000 -0.000
0.000 -0.000 -0.000 -0.000 1.000
spin 2
eigenvalues:
-0.000 -0.000 0.000 1.000 1.000
eigenvectors:
0.002 0.004 0.000 0.070 0.924
0.569 0.160 0.006 0.259 0.007
0.032 0.193 0.732 0.040 0.004
0.025 0.615 0.261 0.080 0.018
0.372 0.028 0.001 0.552 0.048
occupations:
0.994 0.056 -0.005 -0.054 -0.014
0.056 0.265 0.107 0.155 0.395
-0.005 0.107 0.044 0.065 0.162
-0.054 0.155 0.065 0.098 0.239
-0.014 0.395 0.162 0.239 0.599
atomic mag. moment = 3.000000
N of occupied +U levels = 14.000000
--- exit write_ns —
_______________________________________________
Pw_forum mailing list
Pw_forum at pwscf.org<mailto:Pw_forum at pwscf.org>
http://pwscf.org/mailman/listinfo/pw_forum
************************************************
Matteo Cococcioni
Theory and Simulation of Materials
École Polytechnique Fédérale de Lausanne
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20180202/a606735e/attachment.html>
More information about the users
mailing list